A Practical Comparison Between Thesaurus and Ontology Techniques As a Basis for Search Improvement

Abstract

ABSTRACT To compare the capabilities of a thesaurus and an ontology, we explored a number of differences between them, employing two specific examples: the National Agricultural Library Thesaurus (NALT) and a crop-pest ontology. For each, we compared the richness of representation of domain knowledge and the capacity for reasoning that could potentially lead to improved ability to retrieve documents, including agricultural publications. Fundamental differences of domain knowledge representation between them were then identified: Formality of language in the crop-pest ontology, logical consistency of concepts and relationships in the crop-pest ontology, and ambiguity of relationships among terms in the NALT. Relationships such as broader term (BT) and narrower term (NT) in the NALT could support a capacity for reasoning based on generalization and specialization, assuming the relationships themselves are valid. However, the crop-pest ontology supports the deduction of conclusions based on domain knowledge described in the ontology, the search for information resulting from logical inference, and the automated validation of logical consistency. We conclude that an ontology can provide a better representation of domain knowledge and more advanced power of reasoning based on the underlying knowledge representation, which could improve searching in agricultural publications.